This section focuses on the bilateral aid, trying to disentangle the factors that determine the donors’ decision of GPG financing. The empirical literature on the factors influencing aid allocation16 provides two patterns of modeling: i) hybrid models and ii) donor interest/ recipient needs models. In hybrid models the aid flows are explained by a combination of variables representing political, commercial, and humanitarian motives. In donor interest models and in recipient needs models, the egoistic and altruistic side of the action of the donor are separated17. We adapt the hybrid model, focusing on the donor, separating the two stages, how much to give and how to allocate, and concentrating on the first one. The determinants of the financing flows are then to be found not just in the conditions of the recipient and among the hidden interests of donors, but also in the preferences and conditions of the donors18. The result is that recipients are considered as a unicum, which is consistent with our interest in global goods financing, where benefits of GPGs are potentially equal for all countries and the relevant role is that of the financer/producer.
5.1 Determinants of bilateral ODA
As in Reisen et al. (2004), we first analyze the source of GPGs financing, namely bilateral ODA, in total and as corrected by debt forgiveness. The variables considered can be roughly divided into two groups, i) indicators of preferences for bilateral aid and GPGs, and ii) indicators of constraints. The first group includes variables that summarizes the country’s position with respect to: (a) the financing of public expenditure; (b) the degree of openness to the rest of the world; (c) the preference for redistribution both within the country and (d) between countries; (e) the importance given to country-specific gains from aid. The second group includes variables related to the state of the donor’s public finances. We expect that when the budgetary situation is under strain, less effort can be devoted to financing development abroad: ODA becomes one of the first items to be cut under a budget tightening, like public investment, and it is resumed when the state of public finances improves.
Table 9 presents the correlations between determinant variables, and the ratios ODA/GDP and ODA transfers/GDP, in the three sub-periods. There are other country-specific factors impact on the level of bilateral aid that we do not explicitly take into account. For example, some countries like Italy prefer to contribute to multilateral agencies rather than to bilateral aid, given the fact that multilateral programs are less labor-intensive and the structural
16 For a survey see McGillivray and White (1993).
17 The types of models appeared in literature are more than just the two types considered here: there are also bias models, stressing certain phenomena like the small-country effect or medium-income effect; developmental models, which, in a certain way, are similar to recipient needs model but with more stress on the role of developmental variables representing the ability to absorb aid; limited dependent variable models including the eligibility for aid choice.
18 Reisen et al. (2004) adopt this interpretation, when correlating GPG/ODA and ODA/GDP with variables
and operational deficiencies of the departments in charge of bilateral flows (Maurini and Settimo, 2009).
We don’t find any remarkable difference in the correlation strength for the two definitions of aid (ODA and ODAT), probably because the reasons for the expenditure are very similar. The three sub-periods present a certain stability in the significant variables, as if the variability of the composition of the aid could be explained by a constant set of reasons. In particular, among the variables related to openness, a large outward direct investment position (as in Reisen et al., 2004) and foreign direct investment outflow are significant and positively correlated with aid for all the sub- periods.
Table 9 - Correlation coefficients for the ODA to GDP ratio
1995-1998° 1999-2002 2003-2006
ODA/GDP ODA
transfers/GDP ODA/GDP ODA
transfers/GDP ODA/GDP ODA transfers/GDP Openness to rest of the world
FDI outflow %GDP- 2006 0.4265**
(0.0878)
Altruism within the country
Social expenditure %GDP 0.2425 (0.3322 )
financial balance (% GDP)
0.4964* Maastricht debt/GDP (a) 0.2238
(0.4845) Source: Author’s calculations based on OECD-CRS data.
Notes: As in Reisen et al. (2004) we report the Spearman correlation coefficients for period averages. * Correlation is significant at a 5 per cent level ** Correlation is significant at a 10 per cent level. ° Luxembourg, New Zealand, Ireland, Greece are not considered because of insufficient data for that period. °° Greece and New Zealand are not considered because of insufficient data. (a) EU countries only; (c) When GPGs are correlated with ODA
The preference for domestic inequality, as summarized by the Gini index, is negatively related to aid financing, meaning that countries that do not allow for much domestic redistribution are less involved in international redistribution. With regard to altruism in international relationships, a larger share of tied aid19 is negatively and significantly correlated with aid giving in the last interval, 2003-2006. It could be that conditionality on aid is associated with weaker altruism and smaller aid flows. Alternatively. the explicit decision to finance more extensively international programs possibly goes hand in hand with the lower necessity to buy internal consensus by imposing conditionality on aid. Moreover, the concern for development and for better quality in the relationships with other countries, as captured by the Commitment to Development Index20, is significantly associated with larger aid giving. Another variable related to altruism in international relationships, Effectiveness of national aid, expressing a great support to aid by the vast majority of the EU citizens (above 70% on average since 1990s), is significantly and positively correlated with ODA21.
Out of the indicators of preferences for public goods, all variables, except expenditure on health, are significantly related to aid. As in Reisen et al. (2004), this supports the hypothesis that a larger government is associated with higher spending also on international programs.
Among the indicators for the state of public finances, the general government financial balance, the Maastricht debt (last two periods), and the interest expenditure (last period and only for ODA transfers) are significant22. This supports the hypothesis that part of the generosity in aid financing is explained by the availability of public saving: countries undergoing a period of public finance distress or reform tend to cut all more flexible budget items, including the support to international programs. This conclusion is shared also by te Velde et al. (2002).
5.2 Determinants of GPGs
The analysis above is performed for the three aggregates of GPGs, adding two variables, population and gross national income pro capita23, as representatives of the potential direct benefits from GPG provision. As in Barrett (2007), we expect that a larger income and a larger
19 Tied aid is defined as loans and grants which are tied to procurement of goods and services from the donor country and from a restricted number of countries. The literature estimates that tying raises the cost of aid projects a typical 15–30 percent and reduces the value of aid by 13–23 percent.
20 There are no data of the CDI for the first two sub-periods.
21 Hudson and Van Heerde (2009), over the period 1990-2007, consider both a strict (including only questions on development aid) and a relaxed (including even questions on poverty in general) measure of public support: they find non-significant relations (respectively a negative and a positive one) with ODA. The authors stress how, even if there is a sort of unanimous consent to public intervention, which is confirmed for national policies, it is less evident for foreign policies. Their finding suggests us to take both results with caution, considering the low level of information of the public and the vagueness and not explicitness of the surveys.
22 General government gross financial liabilities are not significantly correlated with aid, probably because gross debt data are not always comparable across countries due to different definitions or treatment of debt components. In particular, debt data include the funded portion of government employee pension liabilities in some OECD countries, including Australia and the United States. The debt position of these countries is thus overstated relative to countries that have large unfunded liabilities for such pensions, which according to ESA95 are not counted in the debt figures.
23 “This is because people benefit from the supply of global public goods, and their willingness to pay for provision—a measure of their benefit—while not determined by their income, will almost certainly increase in the
number of potential beneficiaries increase the willingness to take part in GPG financing (Table 10).
Table 10 - Correlation coefficients for the GPG ratios
Spearman correlation
1995-1998° 1999-2002°° 2003-2006
Outward direct
Opennessto the rest of the world
FDI outflow as a %
Preference for public goods
Gross domestic
Source: Author’s calculations based on OECD-CRS data.
Notes. ° Luxembourg, New Zeland, Ireland, Greece are not considered because of insufficient data for that period. °° Greece and New Zeland are not considered because of insufficient data. °°° Spain, Portugal are not considered because of insuficient data. (a) EU countries only; * significant at 5% ** significant at 10%
Table 10 shows that the strategic interest component continues to be supported by the statistical significance of the variables included under the heading Openness to the rest of the world. The significance of the Commitment to Development Index confirms the idea that not only selfish interests move developed countries. Preference for national public goods continues to be significantly and positively correlated with GPGs (E aggregate), supporting the hypothesis that countries interested in national public goods are more in favor of GPG financing. This result is different from that found by Reisen et al. (2004, p. 25), where these variables are significantly correlated only with ODA/GDP. Our finding could be due to the overlapping role of GPGs and ODA in certain sectors.
The variables related to potential benefits from GPG financing, are significant but in different sub-periods: population is significant in the first period and per capita GNI in the following two periods.
The state of the public finances continues to be significantly related to GPGs financing.
As expected, the relationship is positive for the budget balance, as in Reisen et al. (2004), and negative for the General government gross financial liabilities, the share of Maastricht debt and interest expenditure on GDP for the EU countries.
In conclusion, the comparison among the three aggregates of GPGs adopted shows the light superiority of the enlarged definition in better catching the different motives behind the choice of GPGs financing.